Regularity, as the number of rules a system rely on, is not the only variable, and not even the effectively most significant one, when it comes to language ergonomic. Just like terseness, it's an easy to measure and easy to make low in ways that are actually working against ergonomy.
If you look at Justine post, Ruby is given as the hardest to make highlighted correctly. Much of it I guess however pertains to the many ways it's possible to switch to in place string feeding. It's harder to deal with for automated parsers, but it's generally not pushing human writers to encode in a way that is later harder to interpret by fellow humans with many surprising pitfalls to keep in mind. It doesn't mean that Ruby doesn't have any pitfall, but at least its community is using this low level of suprise as a guiding principle.
Optimizing for the shortest encoding string is missing the point that the biggest cost will be how much time some human being will need to understand correctly what was the intended result. Python used to emphase easy to read over syntax tersrness golf. Looks like it changed.
Now, they are people who love that kind of golf, I'm sure. So for them, that's certainly good that Python is digging that rabbit hole.
It's ok not everyone share the same taste, isn't it?
Hell is paved with good will I guess. Probably Justine should update https://justine.lol/lex/